Traits unification #1625
Traits unification #1625
Conversation
Co-authored-by: chriseth <[email protected]>
|
Just saw that side_effect_check also needs to take the impls into account. Maybe there is some code to be shared between side_effect_check and the trait checks? |
pil-analyzer/src/pil_analyzer.rs
Outdated
| // We filter out enum type declarations (the constructor functions have been added | ||
| // by the statement processor already). | ||
| // For Arrays, we also collect the inner expressions and expect them to be field elements. | ||
|
|
||
| for (name, trait_impls) in self.implementations.iter_mut() { | ||
| let (_, def) = self | ||
| .definitions | ||
| .get(name) | ||
| .get_mut(name) |
There was a problem hiding this comment.
Why do you need a mutable reference here?
There was a problem hiding this comment.
The generic type that I am going to “specialize” using the implementation, is stored in the trait declaration, that's why the mutability.
pil-analyzer/src/traits_resolver.rs
Outdated
| .function_by_name_mut(&named_expr.name) | ||
| .expect("Function not found in trait declaration"); | ||
|
|
||
| trait_fn.ty.substitute_type_vars(&type_var_mapping); |
There was a problem hiding this comment.
Ah, now I see! Does it even have a type? Shouldn't we just take the type from the trait and then let the type inferenec algorithm do its job?
There was a problem hiding this comment.
I.e. just return trait_type.clone().substitute_type_vars(&type_var_mapping)?
There was a problem hiding this comment.
And then I think the type inference algorithm will store the type anyway at the end.
ast/src/parsed/mod.rs
Outdated
| panic!("Expected trait declaration"); | ||
| }; | ||
|
|
||
| pub fn specialize_trait_type(&self, trait_decl: &TraitDeclaration, fn_name: &str) -> Type { |
There was a problem hiding this comment.
Nice! And now I think we could rename this to type_of_function.
pil-analyzer/src/pil_analyzer.rs
Outdated
| @@ -220,7 +220,9 @@ impl PILAnalyzer { | |||
| for impl_ in impls { | |||
There was a problem hiding this comment.
Is this inside the loop over the definitions?
No description provided.